Community change is one of the few constants in nature, and the balance of mechanisms influencing this change is central to understanding the structure and functioning of communities and ecosystems. Newly established communities undergo succession and can change in diversity and composition as local environmental change, interspecific interactions, and immigration play out over time. Understanding the influence of initial conditions and priority effects (long-term consequences of the initial community composition and species identity) on community change is critically important for both evaluating ecological theory and predicting restoration outcomes. 

Here I evaluate how initial experimental conditions in 2012, such as initial sown species richness, phylogenetic diversity, and early biomass production, along with priority effects caused by the identity of sown species, influence subsequent plant community composition and the number of colonizing species after nine years of uninterrupted natural colonization.  

I found that sown phylogenetic diversity (measured as mean pairwise distance) indirectly affected the number of colonizing species by increasing biomass production early on and plots with more biomass in 2012 were colonized by fewer species. Individual species influenced the number of colonizing species with taller species reducing the number of colonists and some shorter species increasing the number of colonists. 

Synthesis: Taken together, these results indicate that initial composition influences the number of colonizing species via community-wide competition. These findings suggest that restoration outcomes can be greatly influenced by decisions about sown species composition and early management practices.

This experiment was conducted in an old field at the University of Toronto’s Koffler Scientific Reserve, located about 50 km away from Toronto, Ontario, Canada (44o02’ N, 79o31’ W), as is detailed in Cadotte (2013), but I supply important details here. In fall 2009 and early spring 2010, a 30 x 30 m field was plowed and disked, and then 100 2 x 2 m plots were set up and separated by pathways 1 m in width. 

In May 2010, the plots were seeded with 1, 2, and 4 plant species from a pool of 15 species, including: Andropogon gerardii (Vitman, Poaceae); Schizachyrium scoparium ([Michx.] Nash, Poaceae); Elymus canadensis (L., Poaceae); E. trachycaulus ([Link) Gould ex Shinners, Poaceae); Asclepias tuberosa (L., Asclepiadaceae); Rudbeckia hirta (L. Asteraceae); Solidago altissima (L., Asteraceae); Solidago nemoralis (Aiton, Asteraceae); Desmodium canadense (L., Fabaceae); Lespedeza capitata (Michx., Fabaceae); Monarda fistulosa (L., Lamiaceae); Pycnanthemum tenuifolium (Schrad., Lamiaceae); P. virginianum ([L.] T. Dur & B. D. Jacks. Ex B. L. Rob & Fernald, Lamiaceae); Penstemon digitalis (Nutt. Ex Sims, Scrophulariaceae); and P. hirsutus (L. Willd., Scrophulariaceae). Note that subsequent analyses refer to 14 species as the two Elymus species were combined because many unreproductive individuals resulted in uncertain identification. Each plot was sown with 3,000-4,000 seeds, approximating natural seed fall rates, divided amongst the number of species in a plot. For further details on seed sources and germination, see Cadotte (2013). Due to a lack of germination of some species, only 89 plots are included in subsequent analyses, and some 4-species plots became 3-species plots. In total, there are 44 monoculture, 25 two-species, 5 three-species, and 15 four-species plots analyzed. See Cadotte (2013) for full details.

All species monocultures were replicated three times. The 2- and 4-species treatments were fully crossed with three phylogenetic diversity treatments: 1) small phylogenetic distances, where plots were comprised of closely related species; 2) large phylogenetic distances, with plots members being distantly related; and finally, 3) medium phylogenetic distances, where plots contain moderately distantly related species, or a mix of close and distant relatives. Since there were fewer possible combinations for the small and large phylogenetic distance treatments, these were replicated 7 times each, for both the 2 and 4 species plots. The medium treatments were replicated 9 times for both 2 and 4 species. The phylogeny was estimated using five commonly sequenced genes: matK, rbcl, ITS1, ITS2 and 5.8s. For details on the phylogenetic inference and the phylogeny used to estimate phylogenetic distances, see Cadotte (2013).

In late July 2012, a single 0.1 x 1 m quadrat was placed in each plot, 0.25 m away from any plot edge. All aboveground biomass was removed, with stems clipped just above the soil surface. Biomass was sorted into constituent species as well as litter (i.e., dead or dying material not connected to a living stem). All samples were dried in a VWR drying oven at 50o C for two days, and then weighed.

Following the 2012 sampling, the plots were no longer weeded or maintained in any way other than leaving the deer fencing in place. To assess colonization of species, the entire 2 x 2 m plots were assessed for species composition and percent cover in late July 2021. In addition to the original experimental plots, I also sampled 10 ‘control’ plots. These plots surrounded the experiment and were plowed as part of the set-up for the experiment. Thus, these plots represent natural colonization and succession without any effect of sown species.